Newsgroups: comp.robotics
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!howland.reston.ans.net!ix.netcom.com!netcom.com!nagle
From: nagle@netcom.com (John Nagle)
Subject: Re: Biped robot designs & ideas (Question)
Message-ID: <nagleD3A9G8.M37@netcom.com>
Organization: NETCOM On-line Communication Services (408 261-4700 guest)
References: <3evm76$84n@newsbf02.news.aol.com> <3g6ehl$j7j@hpuerci.atl.hp.com> <D31ML8.AuE@armory.com> <nagleD34pMu.CzD@netcom.com> <D3A0Fy.CMz@armory.com>
Date: Tue, 31 Jan 1995 18:45:44 GMT
Lines: 41

rstevew@armory.com (Richard Steven Walz) writes:
>In article <nagleD34pMu.CzD@netcom.com>, John Nagle <nagle@netcom.com> wrote:
>>     Actually, balance is hard, but not that computionally intensive.
>>See Raibert's book.  
>>
>>     I've recently figured out how to extend automatic balance control
>>to slippery surfaces and hills.  The patent is pending.  The paper will
>>appear in IEEE Computer Animation '95.  Again, it's subtle, but not
>>that compute-intensive.  
>>					John Nagle
>------------------------------
>Many congratulations then, John, but is it feasible for the hobbyist level
>roboticist? And can you tell us if you used rule based or fuzzy sets? And
>would you tell us if you are going to put the paper on the net and where,
>if you would? Thanks, and kudos to your innovation!

      No rule bases, no fuzzy sets.  It's all straightforward control
theory at the lower levels.  This may change as I make it adaptive.

      So far, all the self-balancing legged machines come from the MIT
Leg Lab.  None of them are "free"; they're all connected by cables and
hoses to external hydraulic and pneumatic supplies.  I think it's
possible to build an all-electric runner using some of the new
high-performance cobalt-neodymium linear actuators, but nobody has
tried it yet.  Prof. Ossana Khatib at Stanford is looking for a grad 
student who wants to try it.

      I do all my legged-running work in simulation, although I've built 
wheeled machines in the past.  I've been using Working Model from
Knowledge Revolution, which is a nice 2D simulator for PCs and Macs.
3D is still a problem, because most of the 3D simulators don't get
friction at foot/ground contact right.  I'm talking to simulator
vendors about fixing this.

      I'm thinking of setting up a WWW page for my papers and videos,
but I haven't done so yet.  If anybody is really interested, I'll fax them
a copy of the paper (an ASCII version loses too much).  But read
"Legged Robots that Balance" by Marc Raibert first, or it won't make
any sense.

						John Nagle
